The Association Between Conventional Risk Factors and Diabetes Is Weak Among Urban Tanzanians

Sub-Saharan Africa is experiencing the highest relative increase in diabetes (1), but data are scarce, with only nine countries reporting prevalence surveys within the last decade (2). While biological and behavioral risk factors for diabetes are fairly well-described in high-income settings, it is not known to what extent such risk factors contribute to the diabetes epidemic in low-resource settings. The aim of the current study was to assess whether conventional Western population risk factors for diabetes are appropriate in identifying individuals with diabetes among urban Tanzanians.

The study took place in the urban setting of Mwanza City, which is located in northwestern Tanzania. The current study was part of a larger nutrition study (2006–2009) (3), and 743 healthy control subjects chosen from among tuberculosis patients were invited to participate. Individuals ≥15 years of age, nonpregnant, and without severe diseases were invited to participate. Ethical permission was obtained from the Medical Research Coordinating Committee of the National Institute for Medical Research, Tanzania. Standardized questionnaires were used to collect data on demography, smoking habits, and alcohol intake. Information on deliberate soil intake (geophagy) and frequency (pooled to any or none) was asked for. Diabetes was diagnosed using a standard oral glucose tolerance test (OGTT). Furthermore, anthropometric measurements and HIV status were assessed. The association between diabetes and known risk factors (e.g., anthropometry, age, sex, smoking, alcohol, and obesity) as well as potential risk factors were examined using logistic regression analyses.

The mean (SD) age was 34.2 years (12.9 years), and 6.5% of participants were >55 years of age. The mean (SD) BMI was 22.5 kg/m² (4.1 kg/m²), with 19.2% being overweight (BMI ≥ 25 kg/m²) and 9.1% being underweight (BMI < 18.5 kg/m²). More men than women consumed alcohol (28.5 vs. 19.4%, P = 0.004) and were current smokers (24.2 vs. 1.9%, P < 0.001). Geophagy was only common among women (21.4 vs. 0.3%, P < 0.001). From the OGTT results, 7.7, 5.4, and 8.8% had impaired fasting glycemia (IFG), impaired glucose tolerance, and diabetes, respectively.

The association between potential risk factors and diabetes is shown in Table 1. While diabetes was not associated with age or anthropometry, it was associated with smoking (OR 2.12 [95% CI 0.89–5.04]), female sex (OR 1.66 [95% CI 0.97–2.84]), and geophagy (OR 2.22 [95% CI 1.06–4.64]). IFG was negatively associated with the intake of alcohol (OR 0.31 [95% CI 0.12–0.81]) and was less common among overweight participants (OR 0.31 [95% CI 0.12–0.82]). Linear models of the association between potential risk factors and blood glucose levels (fasting, post-OGTT) provided similar results (data not shown).

Undiagnosed diabetes seems common among urban Tanzanians, but the association with conventional biological and behavioral risk factors was poor. Interestingly, geophagy, usually a marker of iron deficiency, was a risk factor for diabetes. Most of the individuals with diabetes had a normal BMI, as previously reported from Kenya and Ethiopia (4,5). The underlying mechanism of the diabetes found in this population could be poor β-cell function or hepatic insulin resistance rather than peripheral resistance, which also would explain the high prevalence of IFG. There is a need to identify risk factors for diabetes in resource-constraint settings to develop screening tools and, ultimately, to explore whether such diabetes also leads to later complications.